Travis Ennis (@travisennis) header
Travis Ennis (@travisennis) profile picture

Working on acai and cake, but always looking for new ideas to dig into.

Indiana, USA
Joined June 2007
Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)Photo by Travis Ennis (@travisennis)
Used a prompt from @mattpocockuk to remove comments, and once I removed the comments, I found out that some of the comments were convincing the agent that certain pieces of code were working fine, when they weren’t.
2
24
Spent a lot of time this week removing things that I had grown to rely on to do my agentic software development, only to find that once they were removed, many of them were covering up issues within my system.
14
Decided to wipe out ADRs, ExecPlans and tests from my two biggest repos. Working on ripping out task/issues. Curious what it’s like to work with far fewer things the agent has to deal with while it works.
11
Travis Ennis retweeted
Inkling 2 mini now on MLX-VLM 🚀 github.com/Neywa-Labs/mlx-vl…
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index. thinkingmachines.ai/news/ink…
20
30
3
521
27,129
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index. thinkingmachines.ai/news/ink…
611
1,646
993
17,580
3,066,664
Travis Ennis retweeted
Introducing Qwen3.8-27B-pi-adapted 🤗 We adapted Qwen's model specifically for the pi coding agent using reinforcement learning and a LoRA adapter. The results, measured as pass@1: 🚀 SWE-bench Pro (Hard 51): 29.4% → 62.7% 💻 Terminal-Bench 2.1: 39.5% → 55.8% ⚡ 55–62% fewer tool calls per task This is the power of open models: you can adapt them to YOUR tools and YOUR use cases. Let’s build! 💛 Open weights: huggingface.co/harness-hub/Q… Methodology: huggingface.co/blog/harness-…
308
644
175
5,998
456,765
Introducing Inkling 2 Small. Scoring 31 on the AA Intelligence Index, it's a big jump over its predecessor. Inkling 2 Small is available for fine-tuning and inference on Tinker today, with full weight release planned soon. thinkingmachines.ai/news/ink…
300
596
508
6,457
2,027,604
Today we release Ripple-65B Preview, our 65B dense LLM we pretrained from scratch. Ripple sets a new state-of-the-art for U.S. open-weight models in agentic coding tasks, surpassing significantly larger models including Muse Spark Open-816B, Beam-501B, Inkling-975B, and Nemotron 3 Ultra-550B. Learn more: ricursive.com/ripple
399
544
309
4,544
789,538
I’m not sure they do. It does seem like a waste in some respects.
Do individual limits still make sense? As capabilities increase you want to work with and support an agent that represents your project, that runs loops and evolves your codebase. The humans merely curate the automations.
1
22
There was a lot. As more people get hired and businesses acquired the early energy that was mostly open experimentation in AI agents is gone.
so many acquisition announcements today
1
19
Models priced like Opus, Astra, etc are just not going to be daily drivers for most people in enterprises. Providers who have good cheaper options are going to start to win here.
NEW: Apple is cutting employee AI spending budgets even as it pushes workers to use the technology more. Engineers have seen daily limits cut from $300 to $200, with further reductions to $100 planned. Other teams could see limits fall to $25. bloomberg.com/news/articles/…
17
Travis Ennis retweeted
👨‍💻 THREAD on how data center-related ballot measures are faring tonight data centers are on the ballot in ~40 localities across 13 states, including proposed bans, votes on specific projects, and recalls of pro-data center politicians
85
649
354
3,349
1,070,405
I read this and realized this is not how I view the world and it upset me.
In a weird way, everybody is rooting for you. When you go to a job interview, the interviewer wants you to be amazing so they don’t have to look anymore. when you go on a date, the person wants to have an amazing date with you. If you make a piece of content, people want to see amazing pieces of content and they have no problem if it’s yours. the instant you get into the mindset that people want you to do well, your entire world gets better. And you’ll be shocked to know this pattern keeps repeating in different areas of life.
1
20
Travis Ennis retweeted
New Anthropic research: progress on provable inference for large language models. Here, we introduce FlashProve. Building on established cryptographic techniques, FlashProve allows developers to verify whether a given response came from a given AI model. It’s substantially more efficient than prior open-source methods for doing so. These proofs have a range of applications. They can, for instance, make it harder for attackers to hijack a chat, or enable safety evaluators to verify that the model they’re testing is the same one being served to customers. For our full write-up, see here: anthropic.com/research/prova… Further progress in provable inference will likely be an important part of improving trust in AI deployment. To that end, we're open-sourcing FlashProve’s code, here: github.com/anthropics/FlashP… We’re grateful to our collaborators: Alessandro Chiesa, @GiacomoFenzi, @gwberg95, @miranda__christ, Rahul Dalal (rdalal.github.io) and Ben Mayer (@transmissions11).
655
550
507
5,314
774,412
Travis Ennis retweeted
The code, not the spec, should be the center of your universe
147
53
16
1,275
116,025
US open-weights models made gains in both intelligence and parameter efficiency last week: ➤ Thursday morning: NVIDIA released Nemotron 3.5 Bolt and Nemotron 3.5 Super VL. Super VL scored 26.5 on the Intelligence Index with 120B parameters, pushing Inkling Small off the Pareto frontier ➤ Thursday midday: Meta's Muse Spark Open scored 38.9, more than 12 points ahead of Super VL and the highest intelligence yet for a US open weights model ➤ Friday morning: Beam 1.0, Reflection's first model, scored 30.3 with 501B parameters, joining the frontier between Super VL and Muse Spark Open See our full comparison of open weight models on Artificial Analysis artificialanalysis.ai/models…
20
24
2
232
22,552
Travis Ennis retweeted
🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4
1,627
4,121
2,855
37,217
8,449,299
Travis Ennis retweeted
I think I found a planet nobody knew existed. It's been hiding in NASA telescope data for seven years. At the end of this thread, you can fly to it yourself.
306
836
360
8,722
1,220,471
This is what ADRs are for and most models seem to use them pretty effectively.
we've mostly gotten past the slop issue in this new world of working together with AI the next problem is people keep accidentally undoing functionality this is happening because everyone is more in everyone else's business. hit an issue? fix it right away without nagging anyone the problem is it's unclear how intentional the current behavior is and what absolutely should not be rolled back you can say tests should be this. but they don't capture the fact that you discarded seemingly valid solutions 1-4 and settled on 5. someone later (an agent) might thing "oh 3 is simpler" any additional document also doesn't feel like it solves the problem because it's yet another thing that can drift where is the source of truth for how your software is supposed to work?
27